Chapter 12 - modelling Flashcards
What are the two main approaches to modelling for pricing?
- cashflow approach
2. equation of value / formula approach
Models may also be defined by the types of business they are modelling
- single policy profit test model - projects expected cash and profit flows from a single policy from date of issue
- new business model - this projects all the expected cash and profit flows from future sales of new business
- existing business model - this projects all the expected cash and profit flows arising from exiting business at a particular time
- full model office - this is essentially the sum of new business model and existing business model
Main uses of models
- costing and reserving for options
- model office - new business projections, EVs, solvency, takeovers
- reserves - statutory and management accounting
- pricing - profit, premium rates
What is a model point?
A model point is a data record that is fed into the computer as input for the modelling program. It will represent either a policy or group of policies, containing data on the most important characteristics of the policy.
Requirements for a good model
VARIABLE CRISPS CARDS + EEL
- valid
- adequately documented
- rigorous for its purpose. produces realistic results for a wide range of circumstances
- input parameter values are appropriate and includes all necessary input parameters to reflect the characteristics of the product
- assumptions reasonable
- behaviour reasonable
- length of run time not too long / expense not too high
- easy to understand
- communicable workings and output
- reflects risk profile of product being modelled
- independent verification of output
- sensible joint behaviour of variables
- parameters allow for significant features
- simple but retains key features
- clear results
- a range of implementation methods
- refinable and re-developable
- dynamics between assets and liabilities
Requirements for a good health insurance model
CRISSP
- needs to allow for all the cashflows that may arise, which depend on the nature of the contract, in terms of premium and benefit structure and any discretionary benefit such as options to convert, extend or increase cover without evidence of health
- allow for cashflows arising from any supervisory requirement to hold reserves and maintain an adequate margin of solvency
- the model will need to project separately the cashflows arising from different states and reflect the transitions between these states
- the cashflows need to allow for any interactions, particularly where assets and liabilities are being modelled together
- the ability to use stochastic models and simulations needs to be allowed for where appropriate e.g. to simulate claims distribution
Features of deterministic modelling
- each of the parameters has a fixed value
- the model produces the result in the form of a point estimate
- it is possible to sensitivity test the results of a deterministic model by running the model with different parameter values
Features of a stochastic modelling process
- some of the parameters are allowed to vary and have their own distribution functions
- a stochastic model must be run many times using random samples from the distribution functions
- the model produces results in the form of a probability distribution
Why is stochastic modelling more important for healthcare insurance than for pure life insurance?
- With healthcare insurance products, the future incidence experience is far less easy to predict.
- The added difficulty lies with the potential benefit amount, which may vary by policy specified inflation (LTCI), medical inflation (PMI), by changes to medical protocols (PMI) or other factors
- with such uncertainty and hence volatility of cashflows, it is important to be able to project the distribution of possible future outcomes
When is a stochastic model valuable?
- when you are trying to assess the impact of guarantees
- when the variable of interest does have a reasonably stable and predictable probability distribution
- for indicating the effect of year-on-year volatility on risk
- for identifying potentially high risk future scenarios
Disadvantages of stochastic modelling
- time and computing constraints (so stochastic modelling work might be done with a very simplified version of the model)
- the sensitivity of the results to the assumed values of the parameters involved. This can lead to spurious accuracy
Calibration of stochastic models
- risk neutral (market-consistent) calibration
- real world calibration
Market-consistent calibration
- for valuation purposes, usually where options and guarantees are involved.
- aim to replicate the market price of financial instruments
- the idea is that if the model can closely reproduce the market price for quoted assets, then they should be able to closely reproduce the market prices for unquoted assets and liabilities
1. choose a number of financial instruments for which you know the price
2. a model is then built that projects the cashflows of these instruments under a range of scenarios
3. the parameters are then chosen in such a way that the average present value of the cashflows from the modelled simulations is sufficiently close to the known market price
Real world calibration
- usually used for projecting into the future e.g. determining the appropriate amount of capital to hold to ensure solvency under extreme adverse scenarios at a given confidence level.
- focus is to use assumptions that reflect realistic long-term expectations and that consequently also reflect observable real world probabilities and outcomes
- determine model parameters using our expectations of the future
Sensitivity to choice of model point
- if adequate set of model points has been chosen, shouldn’t be necessary to test for model point error
- effect of a different choice should be assessed if less than ideal number of model points has been assumed
- if a large number of runs are required to test thoroughly the various parameter sensitivities, then the model might be recreated with a much smaller number of model points